Construction and Validation of a Prediction Model for Intraoperative Hypoxemia in Elderly Patients Under General Anesthesia Based on Decision Tree

doi:10.12104/j.issn.1674-4748.2025.01.019

Abstract：Objective: To construct and validate a prediction model for intraoperative hypoxemia in elderly patients under general anesthesia based on a decision tree. Methods: A retrospective study was conducted on 132 elderly patients who underwent general anesthesia surgery at the hospital from January 2022 to December 2023. These patients were divided into two groups based on whether they experienced intraoperative hypoxemia: the hypoxemia group (n=26) and the non-hypoxemia group (n=106). Clinical data of both groups were collected and analyzed. Univariate and multivariate analyses were used to identify the influencing factors of intraoperative hypoxemia in patients. A decision tree-based prediction model was constructed based on the results of multivariate analysis. The predictive value of the model was evaluated using the receiver operating characteristic (ROC) curve. Another group of 58 elderly patients who underwent general anesthesia surgery at the hospital from January to March 2024 was selected as the validation group. Results: Multivariate analysis showed that body mass index (BMI), comorbid hypertension, smoking history, preoperative transcutaneous oxygen saturation (SpO2), and surgical duration were independent risk factors for intraoperative hypoxemia in elderly patients under general anesthesia (all P < 0.05). A decision tree prediction model for intraoperative hypoxemia in elderly patients under general anesthesia was constructed based on the multivariate analysis results. The model consisted of three layers and 12 nodes, with surgical duration being the most important predictive factor. ROC analysis showed that the area under the curve (AUC) of the decision tree model in the modeling group was 0.884 [95% CI (0.817, 0.952), P < 0.05], and the AUC of the decision tree model in the validation group was 0.864 [95% CI (0.765, 0.963), P < 0.05]. Conclusion: The decision tree-based prediction model for intraoperative hypoxemia in elderly patients under general anesthesia constructed in this study has good predictive performance and can effectively identify high-risk elderly patients for intraoperative hypoxemia. Clinically, corresponding nursing measures can be formulated, modified, and implemented based on this model to reduce the occurrence of intraoperative hypoxemia.